Financial stability of banks: traditional methods of the Central Bank against self–organizing Kohonen cards

Abstract

The article discusses the problems of assessing the financial stability of commercial banks in the context of financial market turbulence. A comparative analysis of the traditional approach implemented by the Central Bank (a system of supervisory requirements, capital adequacy standards and mandatory reserves) and an alternative method of data mining — Kohonen self–organizing maps (SOM) is carried out.

The author substantiates that the classical methods of the regulator based on strict regulatory values (instruction No. 199–I, etc.), despite the fact that they provide a unified assessment, have limited predictive ability and do not always allow timely detection of latent signs of bank destabilization in the early stages. In contrast, Kohonen’s self–organizing maps offer a tool for nonlinear clustering and visualization of multidimensional data, which makes it possible to identify hidden patterns, «weak signals» of crisis phenomena and assess the stability of a credit institution in dynamics based on neural network modeling.